diff --git a/src/compiler/nir/nir_opcodes.py b/src/compiler/nir/nir_opcodes.py
index ef8de024efb..4f4ee3d6ebc 100644
--- a/src/compiler/nir/nir_opcodes.py
+++ b/src/compiler/nir/nir_opcodes.py
@@ -956,25 +956,10 @@ opcode("fdph_replicated", 0, tfloat, [3, 4], [tfloat, tfloat], False, "",
        "src0.x * src1.x + src0.y * src1.y + src0.z * src1.z + src1.w")
 
 # The C fmin/fmax functions have implementation-defined behaviour for signed
-# zeroes. However, SPIR-V requires:
-#
-#   fmin(-0, +0) = -0
-#   fmax(+0, -0) = +0
-#
-# The NIR opcodes match SPIR-V. Furthermore, the NIR opcodes are commutative, so
-# we must also ensure:
-#
-#   fmin(+0, -0) = -0
-#   fmax(-0, +0) = +0
-#
-# To implement the constant folding, when the sources are equal, we use the
-# min/max of the bit patterns which will order the signed zeroes while
-# preserving all other values.
-for op, macro in [("fmin", "MIN2"), ("fmax", "MAX2")]:
-    binop(op, tfloat, _2src_commutative + associative,
-          "bit_size == 64 ? " +
-          f"(src0 == src1 ? uid({macro}((int64_t)dui(src0), (int64_t)dui(src1))) : {op}(src0, src1)) :"
-          f"(src0 == src1 ? uif({macro}((int32_t)fui(src0), (int32_t)fui(src1))) : {op}f(src0, src1))")
+# zeroes. However, SPIR-V requires IEEE 754-2019 minimumNumber/maximumNumber:
+# -0 compares less than +0.
+binop("fmin", tfloat, _2src_commutative + associative, ("util_min_num(src0, src1)"))
+binop("fmax", tfloat, _2src_commutative + associative, ("util_max_num(src0, src1)"))
 
 binop("imin", tint, _2src_commutative + associative, "MIN2(src0, src1)")
 binop("umin", tuint, _2src_commutative + associative, "MIN2(src0, src1)")